Looks like HP kicked Oracle from the courtroom to the gutter, but somehow still lost the overall war. Oracle set out to gut HP’s high end servers, and did so, the recent court ruling doesn’t change a thing.
If you recall, Oracle publicly and directly kicked Intel’s Itanium in the teeth last year. Instead of the usual tap-dancing, vague references, and implied faults, Oracle said that Itanium was dead, it had no future, and they wouldn’t pour water on one if it was on fire. While the sentiment is hard to argue, HP claimed that Oracle was contractually obligated to keep making Itanium software, and the platform did indeed have a bright future ahead of it.
While we don’t buy HP’s glorious vision any more than we bought Intel’s glorious vision, even if it was spit-shined every time the latest version flopped, HP did have a rather solid contract with Oracle. That said, Itanium did accomplish its goals, and ironically, Oracle was the biggest beneficiary of that collapse. As you might have guessed, it went to court, and today, the court sided with HP.
In duelling press releases, HP puts forth a shining happy vision of the future for Itanium, and the Oracle software on it too. This is said to continue for as long as HP wants to keep pushing Itanium products, and for computers of this nature, that is measured in decades. As far as HP is concerned, they win, Oracle loses, and their customers get everything HP promised they would.
Oracle on the other hand is a bit more demure in its response. HP’s release was titled, “HP Wins Court Ruling in Itanium Litigation”, Oracle’s version is, “Oracle Issues Statement”. When Larry pouts, he sure goes all out, and Oracle is at times nothing more than an extension of his ego, bruised though it may be. Oracle tried to twist the knife in HP’s back, and lost badly. While damages have not been listed yet, most observers, and the author, seem to think it will be measured in billions of dollars.
This is the second time in as many months that Oracle’s courtroom overreaches have seen it slapped down like a petulant child. In the Google vs Oracle Java/Android mess, the billions Oracle asked for turned in to a multi-million dollar bill for Google’s legal fees, but that whole mess has not been fully put to bed yet. Oracle’s track record seems to give vague promise that the US legal system may still function as long as both sides can fund the pursuit.
So, Oracle lost to HP, right? Oracle has basically been required to make everything they promised for Itanium, at no cost to HP, for as long as HP chooses to sell Itanium, plus other yet undefined duties and costs. How can this be considered anything other than a complete HP victory?
Actually, that is easy, because HP lost. The customers for Itanium like to measure uptime in years, and don’t take any crashes, bugs, or problems lightly. If you buy a big Itanium box, speed and performance isn’t your top priority, RAS (Reliability, Availability, Serviceability) is. You want it to not crash, not screw up, and be as fault tolerant as possible. Banks, critical transaction processing, and other non-error tolerant enterprises pay a big premium for this class of server simply to not have problems. You can get faster chips and computers with off the shelf Intel or AMD server CPUs, Itanium is a comparative dog for performance. Then again, where it counts, architectures like Itanium, IBM’s PowerPC, and occasionally a Sun/SPARC chip are worth their weight in gold, nothing else will do.
And on those machines, you run software that is similarly bulletproof, like big databases and transaction processing suites that cost more per core than most cars. And you do so happily because downtime is absurdly expensive, and feet to hold to a fire connected to a throat to strangle if things go wrong is absolutely necessary. For this work, HP, IBM, and a few others provide hardware, Oracle, IBM, and a limited number of players software, and it does deliver what you pay a seemingly silly premium for, RAS.
So HP ‘won’. Oracle has to support Itanium, smile, and do it for as long as HP wants, all at no charge to HP. In short, they have to live up to their contract with HP. What’s the problem? Anyone want to guess how much effort Oracle is going to put in to developing the software for their own in house Sun/SPARC machines vs the ones for HP’s Itanium? On paper, it will of course be completely equal, and no effort will be spared to satisfy Oracle customers regardless of what CPU and platform they choose.
In reality, would you bet your reputation, business, and uptime measured in millions of dollars per hour on the result? Would you trust that Oracle will take the public humiliation being heaped upon them, with good reason, and live up to the spirit of the agreement? They were told to play nice, so they will, right?
Or do you think they will abide by the barest minimum interpretation of the letter of the law? Will Oracle provide whatever minimum they can to keep the wrath of the court mostly away, or will they go all out to do right by HP? Does anyone think bugs will be set upon with vigor by their best programmers, or after a customer can prove that it isn’t HP’s fault, then maybe, when they get around to it, will they hire a few people from the outsourced phone support call center, and train them to look for the bug? After they teach them how to program. If they bother with that bit.
No, HP won the battle, but they lost the war. Oracle has to live up to their contract, and will likely have a long string of onerous technical stipulations to follow, but that is the just letter of the law. Pity the fools that think it will protect them if they use HP Itanium hardware with Oracle software. The second the court’s eyes are off, that ruling won’t be worth the paper it is printed on.
There are bound to be many opinions on this topic, the first one out was David Kanter at Real World Tech. While most have a similar view to the majority of SemiAccurate’s and Kanter’s conclusions, you have to ask yourself, would you bet your company on the resulting promises? If not, HP has conclusively lost the war. It isn’t over yet, but the author can not see any way that HP and Intel can salvage the ashes of the Itanium platform now.S|A
Latest posts by Charlie Demerjian (see all)
- More on Intel’s 10nm process problems - Sep 17, 2018
- Intel puts out another 14nm 2020 server platform - Sep 11, 2018
- Why Can’t Intel Supply Enough 14nm Xeons? - Sep 10, 2018
- Intel can’t supply 14nm Xeons, HPE directly recommends AMD Epyc - Sep 7, 2018
- AMD reintroduces the Athlon name with two CPUs - Sep 6, 2018